Confirming user name and password

[Augie Fackler]

On May 3, 2011, at 4:28 AM, Martin Geisler wrote:
> 
> Benoit Boissinot <bboissin at gmail.com> writes:
> 
>> On Tue, May 3, 2011 at 10:24 AM, Martin Geisler <mg at lazybytes.net> wrote:
>>> Jason Harris <jason at jasonfharris.com> writes:
>>> 
>>>> Hi,
>>>> 
>>>> From MacHg I just discovered the following: I am likely just missing
>>>> something but is there a command which will test to see if a user-name
>>>> + password are acceptable to a server?
>>> 
>>> There is no command that just tests the password -- but you can work
>>> around the problem, see below.
>>> 
>>>> I was using the identify command to see if I could log in with the
>>>> given username and password.
>>>> 
>>>> [Volt:~/test/exclude-changesets] exclude-changesets 3(3) ⌘ hg identify --rev tip https://jfh:dog@bitbucket.org/jfh/exclude-changesets
>>>> 8169bbc96762
>>>> [Volt:~/test/exclude-changesets] exclude-changesets 3(3) ⌘ hg identify --rev tip https://jfh:cat@bitbucket.org/jfh/exclude-changesets
>>>> 8169bbc96762
>>>> 
>>>> Clearly the password can't be both dog and cat. Is it a bug that the
>>>> server doesn't reject the request when the password is incorrect?
>>>> Actually is this bitbucket specific?
>>> 
>>> No, the explanation is much simpler: you are authenticated as needed,
>>> and read-only operations require no authentication on a public repo.
>>> 
>>> You should do a dummy operation that requires authentication. In
>>> Mercurial that means an operation that pushes something to the repo.
>>> Since you don't want to push random changesets around for this check, I
>>> suggest you instead try pushing a new pushkey namespace:
>> 
>> You could theoretically craft an empty bundle to push to the remote server.
> 
> Yeah that would be better since it also works with old servers that know
> nothing about pushkey.

Yeah, the empty bundle is likely better. ISTR having tested that in the past for some reason and it worked fine.